Chaperon and Adjuvant Activity of hsp70: Different Natural Killer Requirement for Cross-Priming of Chaperoned and Bystander Antigens Free

The immunologic function of heat shock proteins (HSP) was initially discovered because of their ability to induce tumor rejection when directly purified from the tumor (1). This activity has then been attributed to their ability to bind cellular peptides and to chaperon them into the antigen presentation pathway of professional antigen-presenting cells (APC; ref. 2) through specific receptors present on these cells (3). Whereas receptors like CD91 (4) and LOX-1 (5) were associated with the cross-presentation of HSP-chaperoned peptides, interaction of purified HSP with TLR4 and/or TLR2 (6, 7) resulted in production of proinflammatory cytokines and up-regulation of costimulatory molecules thus conferring an adjuvant, peptide-independent activity to HSP. Investigations on the innate activity of purified proteins prompted the need to distinguish HSP activities from those of lipopolysaccharide (LPS) that might contaminate the purified chaperon proteins. In fact, removal of LPS from commercially available purified HSP reduced dendritic cell (DC) activation (8, 9), whereas hsp70 and hsp90 have been found to bind LPS (10, 11) thus demanding LPS removal to understand the immunologic properties of HSP preparations. This issue has been approached by searching for differences in the signal transduction pathway of purified HSP and LPS (11) or by genetically modifying the HSP genes to provide a protein either secreted (12) or associated to the membrane of transfected cells (13, 14) thus avoiding the need of protein purification.

We have previously engineered the murine inducible hsp70 into a secreted protein (i.e., hsp70Cκ; ref. 15). Transfected tumor cells constitutively secrete the chaperon protein in the extracellular milieu, where it can be directly available to immunocompetent cells. We have shown that despite relocalization from the cytosol to the secretory pathway, hsp70Cκ retains the ability to interact with antigenic peptides inside the cells and to chaperon them into APC presentation pathway upon secretion. Taking advantage from this system, we have now investigated the peptide-independent adjuvant activity of hsp70. In vitro, secreted hsp70Cκ enhances the uptake of bystander antigenic material and the expression of B7 costimulatory molecules by splenic DC. In vivo, addition of bystander cells secreting hsp70Cκ to cells providing the tumor-associated antigen (TAA) allows a strong activation of the adaptive immune response against the TAA not directly chaperoned by hsp70. To exert its bystander activity, hsp70 requires the presence of natural killer (NK) cells in the periphery, whereas NK cells are dispensable for cross-priming of chaperoned antigen. In sum, whereas chaperon and bystander activities of hsp70 seem equally effective in favoring T-cell proliferation and CTL induction, coupling both activities in the same cell provides the best therapeutic effect induced by tumor cell–based vaccine.

Female C57BL/6×BALB/c (H-2^bxd) and BALB/c (H-2^d) mice were obtained from Charles River Laboratories (Calco, Italy) and used at 8 to 10 weeks of age. OT-I mice are transgenic for the TCR recognizing the MHC class I–restricted OVA peptide. Mice were maintained at the Istituto Nazionale per lo Studio e la Cura dei Tumori under standard conditions according to institutional guidelines.

C26 is a murine colon adenocarcinoma whose immunodominant epitope is the AH1 peptide derived from the envelope protein of an endogenous retrovirus (16); C26αFR is its derivative transfected with the α folate receptor (αFR) model antigen. F1 is a transformed fibroblast cell line that developed spontaneously from BALB/c fibroblasts and that does not express known antigens. F1 transfectants with α-folate receptor (αFR) and/or with a secretory form of murine inducible hsp70 (hsp70Cκ) were previously described (15). Tumor cell lines were cultured in DMEM (Life Technologies, Paisley, United Kingdom) supplemented with 10% fetal bovine serum (Whittaker Bioproducts, Walkersville, MD), 100 units/mL of penicillin, 100 units/mL of streptomycin, and 2 mmol/L l-glutamine at 37°C in a 5% CO₂ atmosphere.

Peptides were synthesized by PRIMM/MWG (Milan, Italy). SIINFEKL is the H-2K^b-restricted epitope of OVA protein recognized by OT-I lymphocytes; SPSYVYHQF (AH1) is the H-2L^d-restricted immunodominant epitope of the C26 env protein. FYFPTPTVL (K_180-187^d) and AGPWAAWPF (D_236-243^d) are the two epitope of the αFR antigen restricted to H-2K^d and H-2D^d molecule, respectively.¹

TMβ1 (anti-IL2Rβ) antibody was purified from the hybridoma ascites and used at 0.5 mg per mice to deplete NK cells in vivo. Effective NK depletion was confirmed by fluorescence-activated cell sorting (FACS) analysis of DX5/CD11b double-positive cells in blood sample (17) and by testing YAC-1 cell lysis by interleukin 2 (IL-2)–stimulated splenocytes.

Antibodies to costimulatory molecule and CD antigens conjugated to FITC, PE, or biotin were from eBioscience (San Diego, CA); FITC anti-DX5 antibody was from Caltag (Burlingame, CA), whereas streptavidin PerCP and PE anti-Vβ5.2 were from PharMingen (BD Biosciences, San Diego, CA). Before staining, cells were incubated with 2.4G2 hybridoma (anti-CD16/CD36 antibody) supernatant to block unspecific binding of antibody to FcR. Data were acquired on a FACScan instrument and analyzed using CellQuest (both from BD Biosciences).

DC were purified from spleen digested for 45 minutes at 37°in Collagenase D (10 μg/mL, Roche, Penzberg, Germany) using anti-CD11c microbeads (Miltenyi Biotech, Bergisch Gladbach, Germany) following manufacturer's instruction. Purified cells were >95% CD11c⁺, as evaluated by FACS analysis.

Naive OT-I T lymphocytes were purified from OT-I spleen. After mechanical disruption, CD8⁺ T lymphocytes were purified using CD8⁺ microbeads (Miltenyi Biotech). FACS analysis revealed that >90% of purified CD8⁺ T cells had a naive phenotype, as evaluated by high CD62L expression.

Natural killer cells for in vivo reconstitution experiments were purified from BALB/c splenocytes using the NK purification kit (Miltenyi Biotech).

Maturation assay. Splenic DC (5 × 10⁵) were incubated at 37°for 20 hours in the culture media derived from F1 or F1hsp70Cκ cell. After extensive washes, cells were double stained with antibodies against CD11c and CD40, CD80, or CD86.

Phagocytosis assay. Splenic DC (10⁵) were incubated with fluorescent tracer in the presence of supernatant from F1 or F1hsp70Cκ cells. Fluorescent latex beads (Polyscience, Warrington, PA) were added for 1 hour to DC at a beads/DC ratio of 100:1, whereas CFSE-labeled irradiated (15,000 rad) C26 cells were added to the DC coculture for 20 hours at a C26/DC ratio of 1:1. At the end of the incubation period, cells were extensively washed and stained with CD11c-PE antibody. A minimum of 10,000 CD11c⁺ events were acquired on a FACScan instrument.

Proliferation of OT-I lymphocytes. DC were stimulated with F1- or F1hsp70Cκ-supernatant for 20 hours, washed, and irradiated (3,000 rad). Titrated amount of DC were added to 10⁵ purified OT-I T cells whose proliferation was evaluate after 3 days by an 18-hour pulse with 1 μCi per well of ³H-thymidine (Perkin-Elmer, Wellesley, MA). The OVA antigen was provided as soluble protein (100 μg/mL) during the 20-hour incubation of DC with hsp70Cκ or pulsed as a peptide (10 μg per 10⁶ DC) just before addition of T lymphocytes to DC. In blocking experiments, the rat anti-mouse CD86 (clone P03.1, eBioscience) or the isotype control antibodies were added to the DC-T lymphocyte culture at a concentration of 15 μg/mL.

BALB/c mice were injected into the footpad with 5 × 10⁶ irradiated C26 cells mixed with equal number of irradiated F1 or F1hsp70Cκ cells. After 3 days, popliteal lymph nodes were digested into Collagenase D solution, extensively washed, and stained with antibodies for FACS analysis of DC phenotype. To detect tumor uptake, C26 cells were labeled with CFSE dyes following manufacturer's instruction.

OT-I lymphocytes purified as described above were stained with CFSE and 2 × 10⁶ labeled cells were transferred i.v. into (C57BL/6×BALB/c)F1 mice. Transferred mice were then injected into the footpad with 100 ng OVA protein mixed with 10⁵ irradiated F1 or F1hsp70Cκ cells. After 3 days, popliteal lymph nodes were removed, minced, and cells were stained with the anti-Vβ5.2-PE antibody specific for the TCR of OT-I lymphocytes and acquired on a FACScan to evaluate dye dilution.

BALB/c mice were inoculated into the footpad with 5 × 10⁶ irradiated C26 cells together with an equal number of irradiated hsp70Cκ-secreting or parental F1αFR cells. After 5 days, lymphocytes from draining lymph nodes were restimulated with irradiated C26 and F1αFR tumor cells and 20 units/mL of rIL2 in complete RPMI medium. After 5 days, cytotoxic activity was tested in a standard 4-hour ⁵¹Cr release assay against concanavalin A–stimulated splenocytes pulsed or not with 1 μg of specific peptides.

Percentage of specific lysis was calculated as 100 × (experimental release − spontaneous release) / (maximum release − spontaneous release), where spontaneous release (never exceeding 10%) was obtained from target cells incubated in medium alone and maximum release from incubation in 1% NP40.

To induce lung metastases, BALB/c mice were injected i.v. with 10⁴ αFR-transfected C26 tumor cells (C26αFR) on day 0. Immunotherapy was done on day +1 +3, +8, and +10 by injection of 2 × 10⁶ of the indicated irradiated cells. PBS was used as control. Mice were sacrificed at day +24 to evaluate the number of metastasis after lung insufflations with 15% India ink and bleaching in Fekete solution.

The significance of the observed effect was valued by t test, except for survival analysis where the log-rank test was used.

Effects of secreted Hsp70 on dendritic cell in vitro. We have previously shown that secreted hsp70Cκ binds to and is internalized by APC in vitro (15). Here, we evaluated the peptide-independent consequences of this interaction by testing whether hsp70 has adjuvant activity in a setting devoid of possible endotoxin contaminations.

To test whether hsp70Cκ induced DC maturation, CD11c⁺ splenic DC were incubated with supernatant from F1 or F1hsp70Cκ cells for 20 hours. In response to secreted hsp70Cκ, DC showed increased expression of CD86 but not of CD40 or CD80 (Fig. 1A).

Because HSP have been described to favor antigen uptake by APC (18), we evaluated whether hsp70Cκ could enhance the internalization of antigens other than those directly chaperoned. Splenic DC were incubated with supernatant from tumors secreting or not hsp70Cκ and with fluorescent latex beads for 1 hour at 37°C and analyzed for beads uptake by FACS. The presence of hsp70Cκ in the culture medium increased the percentage of bead-loaded DC and particularly of DC that have ingested more than one bead (Fig. 1B). This result was confirmed in the context of cellular antigens using irradiated C26 cells labeled with CFSE as source of tumor debris. In the presence of extracellular hsp70Cκ, the number of DC stained positive for CFSE from dying C26 tumor cells increased from 9% to 43% (Fig. 1C).

Such increased uptake correlated with enhanced antigen cross-presentation. Splenic DC incubated with secreted hsp70Cκ and 100 μg/mL of OVA protein for 20 hours were extensively washed, irradiated, and added in titrated amount to purified OVA-specific OT-I transgenic T cells, whose proliferation was measured 72 hours later. The presence of hsp70Cκ during antigen uptake increased the ability of stimulated DC to induce the proliferation of OT-I lymphocytes (Fig. 1D). To test whether the hsp70-enhanced expression of CD86 molecule has a functional significance, splenic DC were incubated for 20 hours with supernatant containing or not hsp70Cκ and then pulsed with the MHC class I–restricted OVA peptide immediately before the addition of naive OT-I lymphocytes in the coculture. In a setting where availability of MHC-I/peptide complexes is not a limiting factor, the increased proliferation of OT-I cells in presence of hsp70Cκ-stimulated DC was due to stronger costimulation through the CD86 molecule and not to MHC class I up-regulation. In fact, the anti-CD86 blocking antibody reduced this proliferation to the background level obtained with mock-treated DC (Fig. 1E), whereas H-2K^b expression on DC was not affected by hsp70 (data not shown).

Secreted hsp70Cκ enhances uptake of dying tumor in vivo. We investigated whether secreted hsp70Cκ improves uptake and processing of nonchaperoned antigens also in vivo. To this purpose, we immunized mice with CFSE-labeled C26 cells mixed with bystander F1 cells secreting or not hsp70Cκ. Three days later, DC from draining lymph node (DLN) were stained for CD11c and analyzed for ingested fluorescent C26 cell debris (Fig. 2A-B). FACS analysis revealed an increased percentage of CFSE-positive DC when immunization was done in the presence of hsp70Cκ, indicating that ingestion of portion of dying cells is favored by secreted hsp70Cκ. Similar to in vitro data, secreted hsp70 augmented both the percentage of phagocytic DC (Fig. 2B) and the amount of engulfed materials (Fig. 2A). No changes were detected in the cellularity and composition of the lymph nodes (data not shown).

Phenotypically, the DC collected from the DLN showed a weak but consistent up-regulation of CD80 and CD86 molecules only in the presence of bystander cells secreting hsp70Cκ (Fig. 2C).

Secreted hsp70Cκ allows a stronger activation of OT-I cells in response to soluble OVA protein. The adjuvant effect of secreted hsp70Cκ on priming of adaptive immune response was tested in vivo using the OVA system. (BALB/cxC57BL/6)F1 mice received CFSE-labeled OT-I T lymphocytes before being immunized into the footpad with 100 ng of OVA protein together with 10⁵ irradiated F1 or F1hsp70Cκ cells. FACS analysis of CFSE dye dilution in DLN collected 3 days after immunization revealed that <35% of OT-I T cells have divided in response to OVA protein given alone (data not shown) or mixed with F1 cells, whereas >86% of them proliferated in response to OVA protein coinjected with F1hsp70Cκ (Fig. 3A-B).

Hsp70Cκ secreted by bystander cells enhance CTL induction against coinjected tumor cells. To evaluate whether the adjuvant activity of hsp70Cκ has a physiologic significance, in vivo, we shifted from the TCR transgenic T-cell transfer setting, where the number of responding T cells is artificially optimized, to naive mice that have a physiologic TCR repertoire. Naive BALB/c mice were immunized into the footpad with C26 colon carcinoma, which expresses the endogenous retroviral env protein as TAA, together with F1 cells expressing the model antigen αFR and secreting or not hsp70Cκ (i.e., F1hsp70CκαFR and F1αFR, respectively).

Lymphocytes from DLN were tested for CTL activity against the nonchaperoned antigen env and the chaperoned antigen αFR. Immunization with C26 admixed with F1hsp70CκαFR cells induced a stronger response against the env-derived peptide than immunization with C26 and F1αFR cell mixture (Fig. 3C). At the same time, the CTL response against the hsp70-chaperoned αFR antigen was greatly enhanced (Fig. 3C; ref. 15). Thus, hsp70Cκ can favor the induction of CTL response against antigens either chaperoned or present nearby.

Natural killer cells are required for the adjuvant activity of hsp70. NK cells have been implicated in the regulation of adaptive immune response via a cross-talk with DC (19, 20). Because we (15) and others (21) have previously shown an interaction between hsp70 and NK cells, we investigated whether NK cells have any role in the adjuvant activity of hsp70Cκ. To this purpose, we used mice depleted of NK cells by means of TMβ1 antibody which targets IL-2Rβ expressing cells (22). We preferred this system to the anti-asialoGM1 polyclonal antibody because it is more specific for NK cells (22) whereas as effective as PK136 (anti-NK1.1) monoclonal antibody in experiments done in (C57BL/6×BALB/c)F1 mice (data not shown). Depletion of NK cells abrogated the hsp70Cκ-increased uptake of CFSE-labeled C26 cells debris by DC (Fig. 4A) and the enhanced priming of CTL against bystander (OVA and env) but not chaperoned (αFR) antigens (Fig. 4B-C).

Natural killer-dendritic cell crosstalk takes place in peripheral tissues. Despite the emerging literature about the interaction between DC and NK cells in vitro (23–25), little is known about such interaction in vivo (26). In the lymph nodes of NK-depleted mice, we found a reduced number of DC that had internalized bystander tumor materials and of CTL primed against the bystander antigen thus suggesting that NK-DC crosstalk in response to secreted hsp70Cκ is an early event that may take place in the periphery rather than in secondary lymphoid organ. To test this hypothesis, we repeated the in vivo uptake experiments in NK-depleted mice by injecting labeled C26 cells and bystander F1 or F1hsp70Cκ with or without purified NK cells. A higher percentage of DC that have internalized labeled C26 cells was found in the DLN of mice receiving hsp70Cκ-secreting cells and NK cells together (Fig. 5A). We confirmed that s.c. injected NK cannot be recovered in DLN (data not shown ref. 27) and concluded that NK cells should be present in peripheral tissue to support the bystander activity of hsp70Cκ on DC. In parallel, NK-depleted mice injected with C26 and F1hsp70Cκ cells together with purified NK rescued the CTL response to the nonchaperoned bystander env antigen (Fig. 5B).

Natural killer cells are required for the therapeutic efficacy of bystander hsp70Cκ. We investigated whether the adjuvant activity of secreted hsp70 could have a therapeutic effect. To this aim, we compared the efficacy of cellular vaccines in which hsp70Cκ is secreted by a bystander cell or by the same cell providing the TAA. Mice were injected i.v. with 10⁴ C26 expressing αFR antigen (C26αFR) to induce lung metastasis and then treated on days +1, +3, +8, and +10, with a cell vaccine based on F1 fibroblasts transfected with αFR and/or secreting hsp70Cκ.

Addition of bystander cells secreting hsp70Cκ to the antigen bearing F1αFR cells (F1αFR + F1hsp70Cκ) reduced the number of lung metastasis (Fig. 6A), but when both chaperon and adjuvant activities were combined in the same cell (i.e., in F1hsp70CκαFR), the vaccine efficacy was greatly enhanced. That is, no lung metastasis were detectable in mice sacrificed at day +24 (Fig. 6A) and 45% of mice were cured over a follow-up of 100 days (data not shown). To analyze the role of NK cells in promoting the therapeutic efficacy of bystander hsp70, we repeated the therapeutic vaccination in NK-depleted mice. As for CTL induction, bystander hsp70Cκ lost therapeutic effect in NK-depleted mice, whereas hsp70Cκ chaperoning the TAA remained effective (Fig. 6B), despite metastasis developed more abundantly in the absence of NK cells (compare Fig. 6A and B).

In this study, we used secreted hsp70Cκ (15) to study how the chaperon and adjuvant activities of hsp70 affect adaptive immune response through DC and NK cells. We found that in addition to chaperoning peptides into DC, hsp70 exerts a peptide-independent activity on DC maturation and function. Splenic DC incubated in vitro with hsp70Cκ, as well as DC from the lymph node draining the site of hsp70Cκ-secreting tumor cells injection, expressed increased level of costimulatory molecules belonging to the B7 family. According to other report (28), DC maturation induced by a single HSP is partial but of functional significance. Indeed, splenic DC incubated in vitro with hsp70Cκ and then pulsed with saturating concentration of OVA-SIINFEKL peptide induced an OVA-specific CD8⁺ T-cell proliferation better than control DC, a results similar to that reported for hsp60 (29). Moreover, experiments done with blocking antibody confirmed the role of CD86 in the enhancement of T-cell proliferation. Thus, our data, in addition to previous reports on engineered gp96 (13, 30), confirm that in settings devoid of possible endotoxin contaminations, HSP have adjuvant activities and thus belong to the endogenous adjuvant proposed by Gallucci et al. (31).

Deeper investigations revealed that the activity of hsp70Cκ was not limited to DC maturation but also favored cross-presentation of “bystander” antigens (i.e., not directly chaperoned by HSP but present nearby). The apparent discrepancy with previous reports indicating that HSP can not induce an immune response against nonchaperoned peptides (32) might be due to the different nature of bystander antigens (i.e., free peptides versus irradiated tumor cells or soluble proteins). Indeed, MacAry et al. (33) showed that an excess of “peptide-free” hsp70 enhances the immunogenicity of suboptimal concentration of peptide-HSP complexes. It has been shown that CMT93 carcinoma cells transfected with a cytosolic form of hsp70 and mixed to B16 melanoma cells do not increase melanoma immunogenicity (18). However, an enhanced immune response against 12B1-D1 leukemia apoptotic cells was obtained by adding liver-purified stress protein (34) thus confirming that the extracellular localization is required to confer adjuvant activity to HSP. TCR transgenic OT-I lymphocytes proliferate in response to OVA protein loaded into cells expressing transmembrane gp96 but not to OVA protein admixed with the same cells (14), suggesting that differences might exist between hsp70 and gp96 in terms of immunostimulatory potency, according to their location in the cell.

In our experimental setting, antigen and HSP have to be given simultaneously, either from the same or from different cells, to induce immune response. We never detected OT-I lymphocyte proliferation or anti-env CTL induction after immunization with bystander cells secreting hsp70Cκ in the absence of antigen (i.e., OVA protein or C26 cells, respectively; data not shown). Thus, CTL against “bystander” antigen should not be confused with a “bystander” activation of T lymphocytes in the absence of TCR triggering due to cytokine production by nearby activated lymphocytes (35, 36). Whereas activation of such bystander lymphocytes occurs with low efficiency and the effects are detectable only in TCR transgenic settings (36), we described induction of CTL against “bystander” antigen in naive mice. Nicchitta's group showed protection against 4T1 mammary carcinoma cells in mice vaccinated with fibroblast secreting gp96 but not sharing any antigen with 4T1 cells (37). They showed that secreted gp96 induced a strong activation of phagocytic cells that in turn generated Th1 polarization of CD4⁺ T cells, with production of large amount of INF-γ (38) but no CTL induction (37).

The activity of secreted hsp70Cκ on bystander antigens is due to the enhancement of DC phagocytosis, as we have shown both in vitro and in vivo. Hsp70 has been previously described to enhance DC antigen uptake without inducing DC maturation (18). Whereas confirming the increased uptake, our data indicate up-regulation of some costimulatory molecule on DC upon encounter with hsp70Cκ. Differences may stem from the origin of DC: bone marrow derived and freshly isolated splenic DC, in other and our case, respectively. Moreover, whereas Vile et al. (18) suggested that hsp70 enhances uptake of tumor lysates by receptor-mediated endocytosis, we have also evidence of enhanced uptake of latex beads, supporting a role for hsp70 in improving both phagocytosis and macropinocytosis, in line with properties reported for gp96 (39). The molecular mechanism responsible for the uptake of bystander antigens by secreted hsp70Cκ can involve both TLR and “endocytic” HSP receptors. Signaling through TLR receptors requires ligand internalization (7) and activates a gene program (40) that temporarily augments phagocytosis by stimulated DC (41). CD91 has been described to induce macropinocytosis of water-soluble dye (42) and internalization of apoptotic bodies associated with collectin receptors (42, 43).

In light of recent reports on the crosstalk between DC and NK cells (44) and on the hsp70-NK interaction (21), we evaluated whether NK cells had any role in the adjuvant activity of hsp70Cκ. NK cells were indeed necessary for the adjuvant but not for the chaperon activity of hsp70Cκ. In NK-depleted mice, the enhanced uptake and CTL response induced by hsp70Cκ against the bystander antigen were lost, and in parallel, therapeutic activity of tumor cell vaccine greatly reduced. This effect is not due to a generalized impairment of CTL induction in the absence of NK cells because the response to the chaperoned antigen was almost unchanged. Our data are in line with Strbo et al. who showed that depletion of NK reduced OT-I T cell expansion in response to cells secreting gp96-Ig and OVA (45).

Such a different NK requirement for CTL induction against bystander and chaperoned antigens suggests that HSP take part to cross-priming in vivo. Despite that several works have described the intracellular pathway involved in cross-presentation of peptides chaperoned by HSP in vitro (46, 47), the in vivo counterpart of this phenomenon has not yet been proved. Studies on the mechanisms of antigen cross-priming in vivo have either negated (48) or supported (49) a role for HSP. Our data showed that hsp70 favors cross priming in two ways, by directly chaperoning antigens into DC and by enhancing antigen uptake and cross-presentation through an NK-dependent mechanism.

We localized this NK-DC crosstalk in the peripheral tissue because injection of NK cells together with tumor cells restored DC uptake of tumor cell debris and CTL induction in NK-depleted mice. Thus, whereas DC have been shown to attract NK into the lymph node, contributing to T-cell activation (27), our data suggest an earlier interaction in the periphery where NK cells take part to the cross-presenting activity of the DC. Search in vitro for the mechanism of hsp70Cκ activity on the DC-NK crosstalk provided unclear results. We evaluated whether hsp70 directly affects NK-mediated cell killing thus increasing antigen availability. Whereas hsp70 on target cell membrane enhanced the cell susceptibility to lysis (15, 50), hsp70Cκ in the cell culture did not affect C26 lysis by IL-2 activated NK (data not shown). We neither found NK cells activation or enhanced DC maturation in vitro when both cells were cocultured in presence of hsp70Cκ (data not shown), as was indeed reported for other maturation stimuli (24). Whether NK cells provide a better antigen source to DC through lysis of tumors or contribute to DC maturation and/or migration to the DLN remains to be elucidated in vivo.

Collectively, we showed that hsp70 has both peptide-dependent and -independent immunostimulatory activities that converge on the induction of an adaptive immune response evaluated as CTL induction and therapeutic efficacy.

However, when both activities are provided by the same cell vaccine, a superior therapeutic effect is obtained thus suggesting that chaperoning of peptides to DC may represent a shortcut for antigen cross-presentation that improves priming of protective immune response.

Grant support: Associazione Italiana per la Ricerca sul Cancro and by Fondo per gli Investimenti della Ricerca di Base.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.